ICC Programming

Hydrological Mapping

4507.1 - High Performance Processing to Enhance the National Hydrography Dataset with Lidar-Derived Drainage Lines

Tuesday, July 4
1:30 PM - 1:50 PM
Location: Maryland B

The National Hydrography Dataset (NHD) of the United States is a comprehensive set of vector features representing the surface-waters in the country, which are used for topographic mapping and scientific investigations. In the conterminous United States, the high-resolution (HR) layer of the NHD is compiled from 1:24,000 or larger scale source material, the detail of which is contingent on the needs of local collaborating agencies. As such, the HR NHD undergoes continuous improvement to furnish more consistent and reliable data. Recent research suggest it is feasible to enhance NHD feature content with drainage lines extracted from lidar-derived digital elevation model (DEM) data (Passalacqua and others, 2012; Poppenga and others, 2010; Poppenga and others, 2013; Stanislawski and others, 2015). These methods could be particularly beneficial in low-relief areas, such as in swampy areas or flood plains, where 10-meter (m) or coarser resolution DEM data do not support the extraction of drainage lines of sufficient quality to enhance the NHD (Stanislawski and others, 2015).

This paper investigates automated methods to extract hydrographic features from lidar-derived DEM data. The U.S. Geological Survey (USGS) has been enhancing its 3D Elevation Program by acquiring 1-m and 3-m resolution DEM data derived from lidar. Because of the large data storage and intensive processing necessary to handle the lidar-derived DEM data, a high-performance computing approach is developed to create a scalable solution to resolve the bottlenecks of computing time and memory. The Broadkill-Smryna subbasin of the NHD is used as a test watershed. This is a low-relief watershed of about 1,970 square kilometers adjacent to the Delaware Bay that requires about 19 Gigabytes of data storage for the 1-m DEM data. The open source Geospatial Data Abstraction Library (GDAL) and the Terrain Analysis Using Digitial Elevation Model (TauDEM) tools are applied through Python programming to extract drainage lines. Methods are implemented on a 10-node Linux high-performance computing cluster.

Results of the extraction of natural drainage lines from 1/3rd arc-second DEM data, nominally 10-m cell resolution, are compared with drainage lines extracted from a 1-m resolution DEM derived from lidar data. The HR NHD flowline network for the test watershed was updated in 2009 with local resolution content derived from 2007 orthophotography and lidar DEM data, and is therefore used as a benchmark for comparisons with the extracted drainage lines. Matching and mismatching features are automatically identified and summarized with the Coefficient of Line Correspondence.

Preliminary analysis indicates drainage lines extracted from 1-m DEM data follow NHD flowlines better than drainage lines extracted from the 1/3rd arc-second DEM data, particularly in the low relief swamp/marsh areas. However, the higher detail of the 1-m data causes some road features to obstruct drainage where a culvert may exist, which forces a few lines to incorrectly follow road edges more so than in the 10-m extracted lines. Processing time for the initial extraction of the drainage lines from the test subbasin took about 6 hours when using 10 processing cores with 128 Gigabytes of memory. Alternative processing methods that allocate additional resources from the cluster and better distribute processing over more cores are being evaluated. Efficient processing methods and workflows that handle the large volumes of high-resolution DEM data and specialized techniques to overcome manmade or other drainage obstructions form a basis for further research and development. With further advancements, it is expected efficient extraction of drainage lines from lidar DEM data will furnish linear feature representations that may either guide updates or be directly included in the HR NHD, specifically in low relief areas, and for improving headwater content.

Lawrence Stanislawski

U.S. Geological Survey

Lawrence Stanislawski is a cartographic research scientist for the Center of Excellence for Geospatial Information Science within the National Geospatial Program of the United States Geological Survey. He studied surveying and forest resource conservation at the Univesity of Florida, and he currently lives in Rolla Missouri. His research interests involve generalization, multiscale representation, geospatial data accuracy, and high performance computing.


Send Email for Lawrence Stanislawski

Barbara P. Buttenfield

Geography Department, University of Colorado-Boulder

Barbara P. Buttenfield is Professor of Geography at the University of Colorado in Boulder. She is Director of the Meridian Lab, a small research facility focusing on visualization and modeling of geographic information and technology. She is a Research Faculty Affiliate with U.S.Geological Survey Center for Excellence in Geospatial Sciences (USGS-CEGIS); a Faculty Affiliate for the Rocky Mountain Census Research Data Center; and she leads the Data Harmonization project for the CU Grand Challenge “Earth Lab” initiative. She publishes research on cartographic generalization, multi-scale geospatial database design, terrain analysis, spatial data integration, and managing uncertainty in environmental models. She is a Past President of the Cartography and Geographic Information Society (CaGIS), a Fellow of the American Congress on Surveying and Mapping (ACSM) and a Fellow of the University Consortium for Geographic Information Science (UCGIS). In 2001, she was named National GIS Educator of the Year by UCGIS, in the inaugural year of the award.


Send Email for Barbara Buttenfield

Yan Liu

Techical Coordinator
Cyberinfrastructure and Geospatial Information Laboratory, University of Illinois at Champaigne-Urbana

Yan Liu is the technical coordinator of the CyberInfrastructure and Geospatial Information Laboratory (CIGI) at the
University of Illinois, Urbana-Champaign.


Send Email for Yan Liu

Jeff Wendel

Computer Scientist
US Geological Survey

Jeff Wendel is a computer scientist with the Center of Excellence for Geospatial Information Science within the National Geospatial Program of the United States Geological Survey. He designs, implements, and administers High Performance Computing clusters and extends Open Source code to run in parallel with the Message Passing Interface (MPI).


Send Email for Jeff Wendel

Paulo Raposo

Assistant Professor of Geography
The University of Tennessee, Knoxville

Dr. Raposo is a cartographer, and an Assistant Professor in Geographic Information Science in the Department of Geography at The University of Tennessee, Knoxville. He received his PhD in Geography, with a specialization in cartography, from Penn State University. His interests are in cartographic design, multiple representation and generalization, geometry, and raster surfaces.


Send Email for Paulo Raposo


4507.1 - High Performance Processing to Enhance the National Hydrography Dataset with Lidar-Derived Drainage Lines


Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for High Performance Processing to Enhance the National Hydrography Dataset with Lidar-Derived Drainage Lines